Science Progress Optimizer

Let All Be Guides! Science Progress Optimizer for Global Optimization

This study developed a novel metaheuristic algorithm named the science progress optimizer (SPO) for global optimization. Inspired by the process of scientific progress, the SPO modelled the facilitating role of radical and conservative innovation. As a swarm intelligence algorithm, the most noteworthy feature of the SPO is that it uses all individuals rather than the best historical individuals to guide the search. We evaluate the SPO and seven recently developed metaheuristics on the CEC2017 benchmark functions (50 dimensions). Furthermore, we evaluate the SPO and five state-of-the-art algorithms on the CEC2017 (100 dimensions) and CEC2022 (20 dimensions) benchmark functions. Finally, the SPO and seven recently developed metaheuristics are used to solve fourteen mechanical engineering problems. The results show that the SPO outperforms seven recently developed metaheuristics in both benchmark tests and engineering problems. Moreover, SPO outperforms the state-of-the-art algorithms on nearly half of the functions and achieves similar results to the state-of-the-art algorithms on many functions.